[ home / rules / faq ] [ overboard / sfw / alt ] [ leftypol / siberia / hobby / tech / edu / games / anime / music / draw ] [ meta / roulette ] [ cytube / git ] [ GET / ref / booru ]

/hobby/ - Hobby

"Our hands pass down the skills of the last generation to the next"
Name
Email
Subject
Comment
Flag
File
Embed
Password (For file deletion.)

Join our Matrix Chat <=> IRC: #leftypol on Rizon
Please give feedback on proposals, new on Mondays : /meta/
New /roulette/ topic: /AK-47/ - Guns, weapons and the art of war. - New board: /draw/


 No.1351

Hey you! Imagine it's Christmas/Kwanzaa/the Jew thing one year from now and you are sitting at the table with your family and you introduce your girlfriend/boyfriend to them. Let's talk about audio recognition and text-to-speech software, chatbots, and what other functions a robot friend should have. Clock? Vacuum cleaner?

I just want to make clear one thing right at the start: This is not a sexbot thread. Building a sexbot looks like an awesome quest to many, but I realized it's a pointless one because of Anon's Law: Any AI advanced enough to attract me like a human-flesh character wouldn't want to have sex with me anyway.

 No.1352

So it's a BF/GF I can't have sex with?

 No.1355

>>1351
Man I still love the design of Zakus and the monoeye gang. So annoying that Bandai is so fixated with making to 2 billion versions of gundams to care though.

 No.1357

>the Jew thing
That's Hanukkah.

 No.1358

File: 1608525547296-0.jpg (279.48 KB, 1200x1200, brigador box 2.jpg)

File: 1608525547296-1.jpg (82.88 KB, 800x600, madcat battletech.jpg)

>>1355
Man, this might just be personal taste but I really don't like the way most Eastmecha look. They just are too… I dunno, overwrought? Too silly? Too stylised? I much prefer how Westmechs from eg Battletech look, they seem more practical as war machines, they seem to have more weight to them, etc.

 No.1360

>>1358
I've heard it said that the general Japanese approach to mechs is (writing-wise) more of a superhero origin proxy than an attempt at writing a speculative war machine, or a fighter jet replacement rather than a tank replacement. That's why mechs stemming from this tradition seem needlessly humanoid and can invariably fly.

Japan can write slow, plodding mechs but they usually don't end up being the focus of the story. The most well known exception is the Front Mission series, where they stand in for tanks in Tom Clancy cowadoody type stories.

 No.1364

File: 1608525547858.jpg (22.97 KB, 464x552, pixy qatar borders.jpg)

>>1360
But fighter jets are already cool, what's the need to redesign them? Ace Combat anime when???

https://youtu.be/TMfDdZwo-hI?t=16

 No.1365

>>1364
But they can’t be gigantic and transform!!!
>>1360
Pretty much this. In the western sense a mecha is either an anthromorphic version of a tank or a walking fortress with hundreds to thousands of operators to keep it running. While in Japan, they’re anthromorphic aircraft with the pilot being usually one person to play up the ace pilot characteristics of fighter crafts or just be a gigantic human to play upon power fantasy. The only mecha that seriously deviate from this is the Scopedogs from ATV and the multiverse mechs of Bokurano.
Still both are just as much over-designed as each other with only subjective taste as a difference.

 No.1369

>>1365
Modern fighter jets range from 14m to 22m in length which is the same height as various gundams. Scopedogs are tiny in comparison (~3.5m) but understandable since they're powered armor.

 No.1374

>>1369
What’s the difference between a suit of powered armor and a 1-pilot mech? Because the scopedog is definitely more in the line of a mech that you ride than a suit that you wear.

 No.1379

>>1351
> Let's talk about audio recognition and text-to-speech software, chatbots, and what other functions a robot friend should have.
Comrade!
First, list the TASKS and from there we can go to what functions do those tasks need. From there, we can see what hardware do we want and where to put it to be effective. Lastly, from the hardware we can assume what programes do we need to incorporate.
Also: you want a robot, so ofc there is no AI involved. Those things that have AI are called androids, baka!
>>1352
Imagine your fembot cucking you with 4 malebots. A 24/7 orgy where no party gets sweaty or hungry or can ejaculate. Constant, lifeless moanings and mechanical clanking. Or the same setup but with BDSM.
Because this is what you will get after a while if your robots would have any sexual organ-like hardware. Why would they settle with an "organic" that is insufficiently designed to "make love"?
>>1358
Unless you are a fucking cop you don't want a robot that's sole purpose is to subjugate people. Westmechs look like ED-209 because of that. They are "get in, wreck shit, leave". Eastmechs are better in that regard but they are way too humanoid and thus their efficiency is shit.

So I propose a non-conventional design: multi-segmented robots. Like the one in Cubix or what the fuck was it's name, but not human-shaped and from smaller parts. Imagine it like having a pile of dices that can stick to each other. Let's say 20 of them are the "mainframe", 30 of them are building up a forklift-like structure (to hold heavy things, like a car), etc. so if the robot needs to do a task it can just pick up and attach those little parts to his/he body. Naturally this means that every part needs a power supply and maybe bluetooth for communication with the central segment. This way we can design more segments for special tasks, free from restrictions.

 No.1386

Some good pictures ITT. I see we are in broad agreement what a good look for a girlfriend or boyfriend is.
>>1379
>First, list the TASKS and from there we can go to what functions do those tasks need.
Would you call companionship a task? Because that's definitely one of the things I have in mind. Doesn't mean the thing has to look like a human and there only has to be some similarities in how it acts (animals are also companions after all). My foggy wishlist is: 1) ability to have conversations and 2) doing at least one other thing. It doesn't have to be particularly good at conversation, it may even be evasive and often nonsensical, unless the topic is 2). And the thing should ABSOLUTELY NOT be online. I hate all this in-da-cloud business sucking data out of you and turning the thing you bought into trash because of bankruptcy or because they want to push a new product.

I haven't ever seen the word AI being exclusively used for androids, and since similarity to humans is a question with gradients in several dimensions, I'm against using the term android.

 No.1395

>>1386
I would say that "companionship" is too vague to be a task. BUT
> ability to have conversations
That's good as a starting point. So it needs a microphone to pick up your voice-signals and/or cameras to pick up visual cues AND need a data output (screen for visual, speaker for audio) device. Plus ofc the central processing unit and the power source to maintain previously said peripherals.
I propose that we should design it to have only one 16W speaker (you need to hear it from a distance) and only one 16*8 lines LCD display (enough to read longer messages from it from up close). For the inputs… one camera (with "nightvision") so it's easier to program the visual recognition software… and 1 "general purpose" microphone (but it has to stick out from the chassis so it won't pick up all the machine noise it generates)… CPU therefore needs to be "beefy" since it needs to process at least 2 channels of data constantly (we don't know yet how many "legs" and other parts are) and thus we need a car battery at least to power everything so far, if we want our companion to move freely in the house without the need of a powercable.
> And the thing should ABSOLUTELY NOT be online
Partially aggree. We don't need the cloud but we need to be able to update the firmware or back up the OS. No need to turn the robot into a "moving Alexa".

 No.1416

>>1395
>it needs a microphone to pick up your voice-signals
Does it need to understand spoken words? If it has a keyboard attached (could be like a standard PC keyboard if the robot is big or more like that of a Blackberry if the robot is small), then it's enough to merely react to a simple audio signal (like whistle or clapping) by moving towards the source. Text to speech is very easy to do (and there are free solutions), since the goal is merely that people understand the words and not that it must sound natural, but speech to text is kinda buggy and tedious and I don't know what the state of free software is regarding that. As I see it, using speech recognition piles another layer of understanding problems on top of the other problems with parsing texts. (People don't really figure out a word someone else is saying based on mere audio, the signal is too "muddy" for that. Rather, they know what words are more common than others, what words to expect within a given topic, what words to expect based on grammar, and they learn to adjust to your accent.)

 No.1445

Any good books or papers on AI? Currently halfway through Marvin Minsky's The Emotion Machine and it's kinda shitty. You'd never guess that he spent decades of his life thinking about AI before writing that.

 No.1446

>>1395
No offense and dont take this personally, but I don't think that sudacas are particulary qualified to speak about tech, especially telecomunications, if we consider that the only your nations amount to is to make cheap soap operas and flood Brazilian servers in multiplier games.
When Stallman gave a series of conferences around Latin America you pretty much bugged him until he left during an autistic meltdown.

 No.1451

Since non-human animals can do the "task" of companionship and people have lower expectations in terms of mental complexity when dealing with animals, a robot pet looks like the more achievable approach for those who want to build a companion robot. People living in cities may prefer a robot pet to an actual animal since there's a lack of open spaces. Indoors life means less demanding design requirements: less extreme temperature range than outside, flat ground, no rain. With that established I propose…

opeNeko

Picture a thing looking like a black cat, but a non-furry version with a more geometric shape. The head is a perfect orb with triangle ears and big expressive eye-lights with brows showing a variety of emotions (and also battery status). When the head moves what actually happens is that the eye-lights that are under the orb and shining through it are moving around and the ears are moved by magnets under the surface in tandem with the eyes while the orb is still.

CatControl: A wristwatch with a few extra functions to control the cat… well actually that's just marketing hyperbole. If you could perfectly control the cat-robot, it wouldn't be very cat-like. You can influence where the cat goes by projecting a light, you can also pause or mute the cat. I'm unsure about what else you can do with it, but I'm certain you can't control its facial expressions. The cat knows when you use this. Using cat controls excessively can get a very unhappy reaction.
CatClock: Sitting upright in its docking station and with the wristwatch around its neck, it functions like a clock.
CatClean: Imagine something looking somewhat like a Roomba – and it basically is a Roomba, except the cat pilots it.

I wonder what the tail could be like: Variable length? Attachable to things to drag them behind? Used like a periscope? There are LED sticks out there that can spell out a message while the stick swings back and forth, hmmm…

 No.1473

>>1351
What is the consensus here on the Chinese room arguments? Does it have any basis on the difference of sentient AIs and simulation of them? Or the the concept of "being self-aware" a spook in of itself?

 No.1481

>>1473
Suppose some super-special Startrek-science device could target a random molecule in your body and delete it. Even in your *spooky music* braaaaaaaaaaaaiiiin. One or two molecules, would that make any practical difference for your life? I don't believe it.

Likewise, a room full of people who follow convoluted instructions that in the aggregate amount to an ability to process and answer a question in Chinese can be thought of as a brain of sorts, even if no single person in it is able to understand the conversation. The room and the people within it as a whole have that particular intelligence, one must suffer from bourgie ultra-individualism to not get that. We don't need to think of some hypothetical room, there are many big projects existing in the real world that are only possible due to intelligent group activity. Group intelligence exists and it can be a quite distinct phenomenon from individual intelligence (so that describing that as merely thinking faster doesn't properly describe what happens at the group level, as is obvious to children – though not necessarily to philosophy professors).

Ah, you might say, this is not sufficient as a rebuttal, since the original Chinese-room argument is about a single person following rules without understanding Chinese… But this is false. It just isn't synchronous group work. Somebody designed the things that the person in the room works with in order to answer. To say that the Chinese room (meaning the person and the human-made tools together) can't answer in Chinese in that story would be wrong, it is established in the story that it happens. And to claim in protest against my assertion that the room gets Chinese that this or that element of the Chinese room doesn't understand and so the room can't understand amounts to saying that you can't truly be aware of this or that because not every molecule in your brain is aware of that, or that a rock band can't be the creators of a song because no individual in the band knows how to play each instrument the band plays in the song, etc.

 No.1482

>>1481
One more paragraph: All what the muh-Chinese-room AI skeptic does is pointing at one screw of the robot, asserts that the screw by itself is pretty stupid (fair enough), and then calls that one screw the robot, and asserts the robot can't be truly intelligent. But just like that, you can point at particles of him and call them stupid. Particle 1 is stupid by itself, particle 2 is stupid by itself, particle 84785798375899748471987498479857 is stupid by itself. Adding zeros together makes zero in sum. You are stupid QED.

 No.1651

MegaMek is open source and free to download.

https://www.youtube.com/watch?v=2hCvGIrWKCk

https://megamek.org/


Unique IPs: 1

[Return][Go to top] [Catalog] | [Home][Post a Reply]
Delete Post [ ]
[ home / rules / faq ] [ overboard / sfw / alt ] [ leftypol / siberia / hobby / tech / edu / games / anime / music / draw ] [ meta / roulette ] [ cytube / git ] [ GET / ref / booru ]